Artificial Intelligence for the Objective Evaluation of Acne Investigator Global Assessment
September 2018 | Volume 17 | Issue 9 | Original Article | 1006 | Copyright © September 2018
Antonella Melina MSc,a Nhan Ngo Dinh MSc,b Benedetta Tafuri MSc,c Giusy Schipani MD,d Steven Nisticò MD,d Carlo Cosentino PhD,a Francesco Amato PhD,a Diane Thiboutot MD,e Andrea Cherubini PhDb
aFaculty of Biomedical Engineering, University “Magna Graecia”, Viale Europa, Catanzaro, Italy bLinkverse, Rome, Italy cInstitute of Molecular Bioimaging and Physiology (CNR-IBFM), Catanzaro, Italy dDepartment of Dermatology, University “Magna Graecia”, Viale Europa, Catanzaro, Italy ePenn State University College of Medicine, Hershey, PA
Introduction: The evaluation of Acne using ordinal scales reflects the clinical perception of severity but has shown low reproducibility both intra- and inter-rater. In this study, we investigated if Artificial Intelligence trained on images of Acne patients could perform acne grading with high accuracy and reliabilities superior to those of expert physicians.
Methods: 479 patients with acne grading ranging from clear to severe and sampled from three ethnic groups participated in this study. Multi-polarization images of facial skin of each patient were acquired from five different angles using the visible spectrum. An Artificial Intelligence was trained using the acquired images to output automatically a measure of Acne severity in the 0-4 numerical range of the Investigator Global Assessment (IGA).
Results: The Artificial Intelligence recognized the IGA of a patient with an accuracy of 0.854 and a correlation between manual and automatized evaluation of r=0.958 (P less than .001).
Discussion: This is the first work where an Artificial Intelligence was able to directly classify acne patients according to an IGA ordinal scale with high accuracy, no human intervention and no need to count lesions.
J Drugs Dermatol. 2018;17(9):1006-1009.
The measurement and grading of acne severity is a recognized challenge1-3, since it is known to depend on the number and type of lesions and the extent of involvement. The method of reference for acne grading is lesion counting4-6; however, it is time-consuming and not well-suited to clinical practice. For this reason, health agencies4,6 recommend the use of an Investigator Global Assessment (IGA) with up to six ordinal grades (clear, almost clear, mild, moderate, severe and very severe). In order to improve reliability, the FDA recommends co-primary endpoints that evaluate both IGA scale and acne lesion count assessments.6 Computational methods based on imaging techniques7-9 have been proposed in scientific literature to reduce the problems associated to low inter- and intra-rater reliability of acne grading.10-12 To date, however, all attempts have focused on the automatization of lesion counting. In the present work, we intend to investigate if an Artificial Intelligence (AI) could perform IGA grading of digital images of acne patients with reliabilities comparable or superior to those of expert physicians.
An imaging device (FaceAtlas, www.linkverse.com) was used to acquire high-resolution multi-polarization images of facial skin using the visible spectrum (Figure 1). During the development of the device, a database of 479 subjects (age range, 11-46 years, median 17.4 years; female 58%) with acne grading ranging from clear to severe, sampled from three ethnic groups (Caucasian-white 79%, African-black 13%, Asian 8%), was acquired at the Department of Dermatology of University “Magna Graecia” after informed consent and approval from the local Institutional Review Board. Three experienced dermatologists assessed acne severity according to the FDA ordinal scale (proportion in the dataset; 0-clear: 18.1%; 1-almost clear: 18.4%; 2-mild: 20.4%; 3-moderate: 23.3%; 4-severe: 19.8%). The clinical evaluation resulting from majority voting among the three raters represented the ground truth for supervised learning.The image acquisition and postprocessing workflow is depicted in Figure 1. As result of this workflow, multiple images collected from a patient were first converted to a standardized planar representation. Subsequently, the planar image was